Thinking about Tracking Design, Part 2: Why track in the first place?

The second part in an ongoing series exploring tracking design

Kanazawa, Japan, April 2016

Hello, and welcome back to Data Rampage!

This is the second part in a six part series that I am doing on tracking design; if you are unfamiliar with the concept of tracking design, I defined it like this in the first part:

Thinking about Tracking Design, Part 1: What is tracking anyways?

Tracking design is the art of designing an effective system for collecting behavioral data from your frontend services. Translated out of data jargon, that just means making decisions about what user actions you want to track on your website and/or mobile app.

Having started the series by discussing what tracking actually is, in this newsletter I’m going to look at the question of why you would want to collect behavioral data in the first place. The following parts will look at what to track and how, and then I will conclude with what you can do with the data you collect from the frontend (spoiler alert: it’s a lot!)

Why do you need to track behavioral data in the first place?

When you set up behavioral tracking software on a digital service, you do so because you want to answer two fundamental questions:

1. Where do your users come from?

2. What do they then do once they arrive?

If you want to build a a successful digital business, these are essential questions to answer; everything else you might want to know flows from them.

As I discussed in my KPI newsletter, every business has larger goals that they want to achieve.

Your business strategy is the goal of your business, and every business has an existential goal.

Here are some examples of goals that different types of businesses might have:

• The goal of a small barbershop might be to provide a decent living standard for the owner as well as a living wage to the employees

• The goal of a software startup might be to grow very quickly and either achieve an IPO or be purchased by a larger firm

• The goal of a privately-held midsize family firm might be to weather business trends and achieve multigenerational wealth and financial stability

• The goal of a major multinational corporation might be to maximize shareholder value via steady growth and high profitability

Well-defined KPIs are how you measure your progress towards achieving your goals, but to measure your KPIs you need to collect data, and the customer-facing side of your services (also known as the 'frontend'), is one of the most important places to collect data, since that is where you can see how customers actually interact with you.

Think of it like this: if you own a store, when customers walk in, do you turn around to face the wall and hope they choose to buy something, or do you pay attention to what they are doing, see what products they look at, see how they move around?

You go for option two, right?

Where do your users come from?

In a recent post I discussed how to do marketing attribution, i.e. how to use behavioral data to understand the impact of different marketing campaigns on the achievement of your business goals.

If you didn't read it before, you can read it here:

How to use Snowplow Analytics for marketing attribution

What I didn't really discuss in that post is the why. Why do you want to know this stuff? How does it help you?

There are a few main reasons:

1. Understanding your audience: By tracking where your users come from, you can gain some useful insights into their interests and behavior that you can translate into ideas for improving your product’s feature set and marketing efforts

2. Measuring the effectiveness of your marketing campaigns: This is the big one; if you’re spending money on marketing, you will want to know if you are getting a return on your investment. Does this individual campaign work? Are search ads more effective on Google or Bing? Should you even bother advertising on social media? Does video advertising drive quality traffics? These are the types of questions that are unanswerable without tracking both where your users come from and what they do once they arrive.

3. Improving your website's SEO (Search Engine Optimization): High-quality traffic that you’ve not paid for? Ideal! Knowing more about search traffic to the site can help you to improve your service’s SEO and bring you more visitors.

4. Identifying potential partnerships: Traffic information can be a very fertile source of new leads for businesses you might want to collaborate with, based on how much traffic they are sending to you (and how good is that traffic).

5. Understand trends over time: Is traffic rising or falling? Are there changes in the composition of your traffic, like you are now getting more organic search traffic but less organic social media traffic?

In general, it’s much easier to get a clear understanding of how your visitors find you on the web than it is on mobile apps. The reason for is that both Apple (iOS) and Google (Android) are extremely parsimonious with the information that they provide about app discovery, so in order to better understand mobile marketing performance you usually need to integrate a mobile attribution tool like Appsflyer or Adjust.

(Note: I should mention that I have much more experience with marketing analytics for the web than for mobile, so if I’m wrong and there are easy mobile attribution solutions, feel free to let me know in the comments!)

On the web it’s much easier, because there is a standard method for extracting referral information from the browser that all trackers use, although with the caveat that you are never able to get referrer information on 100% of your visitors. Why that is the case is a complex matter outside this particular post’s scope, but I recommend reading up on the topic of dark social. The fragility of referral detection is a key reason why I recommend you tag all your marketing efforts, whether performance or content, with identifier tags such as UTM codes, so that you can get a clearer understanding of where your traffic is coming from.

Here’s a practical example to help you understand how this traffic information is essential: let’s say you run an ecommerce store for sneakers and suddenly you have an unexpected burst of sales on a particular item; you normally sell 10 pairs a week, and suddenly you get orders for 100 in a day! What happened? You turn to your traffic data, and you find that most of the new orders came from people who had gone directly to the page via a link on a blog for sneaker collectors. With this information you can reach out to the blog owner and discuss potential business opportunities, for example some kind of affiliate deal or a collaboration.

Without the detailed traffic data, all you would have known was that you had a big spike in orders, and … that’s it! There would be no way to really capitalize on it.

What do your users do once they arrive?

The second key reason that you want to track user behavior is to understand what your visitors do when they are actually using your service. Why is this information useful? A few quick reasons:

1. Understand and improve the user experience: With data about how visitors interact with your service (what pages they see, the journeys they take, what elements they interact with, etc), you can get valuable feedback on your product design decisions

2. Optimization for business outcomes: If you have clear user journeys (like an e-commerce process), event data will help you to pinpoint issues in the funnel, like where users are more likely to drop out of a particular flow

3. Personalization: This is a more advanced use case, but you can use behavioral data to personalize the experience of your service, making it more useful and compelling for visitors. You will have almost certainly encountered this in the form of content recommendation systems, but there are other forms this can take as well

4. Understand customer needs and desires: Behavioral data can inform a wider range of business decisions, for example by providing information about what products are of interest to customers, or what might be new areas for product development

In order to illustrate these points in a more practical way, let’s go back to the same Soundcloud page I introduced in the first part of the series: the listening page for my most recent dj mix, The 90’s Sessions Vol. 2: A Chemical Brothers Special.

If I was the VP of Data at Soundcloud (I did actually interview for that role in the spring of 2021 but they turned me down … oh well!), here are some of the questions I would want to answer when looking at the individual player page:

I’ve annotated the screenshot to show you some of the typical questions that I might ask when thinking about a web page. For example, if I was Soundcloud’s VP of Data, I would be very curious about how users navigate Soundcloud in order to find content; is it an active or a passive process? Are they finding new things to listen to via personalized recommendations or are they actively searching for new content? What can we tell about how well we are surfacing content for them? Are they clicking though tons of stuff or are they listening for a long time to individual tracks? What is their listening behavior like over time? What are the characteristics of particularly engaging content? Is there any relationship between any types of actions and the likelihood they will sign up for Soundcloud’s paid service? If so, what are the key behaviors and can we somehow encourage more users to perform them?

As you can see, there is so much to learn just from a single web page!

To extend this, let me give you an example from my own career. In one of my first data jobs, I was working for a Swiss VOD startup called Viewster (which later pivoted to being a video ad platform called vi and was acquired last year by Outbrain, but that’s a whole separate story), and there was a wide catalogue of movies and tv shows, including a small section of Japanese anime. One of the interesting things that I noticed was that average watch time of anime videos was way above the watch time of the average video on the service, people were repeat watching them more, and they were searching for anime a lot. All of these data points were only available via the frontend, since in the backend we only captured the video plays, and taken together led to us re-focusing the content strategy to be more anime-centric.

So .. that’s it! Hopefully this post will have helped you to better understand why it is beneficial to track user behavior on your website and/or mobile app. Next up I will be discussing what to track, and some of the key questions you need to ask yourself when designing a tracking plan.

One last (music) thing

Yes, as with every post I am ending this one by sharing a dj mix from my very extensive back catalogue. I’ve now been a dj for 26 (!) years, so why not, eh?

For this one, I’m sharing a mix I did last year called Eurotrash 8, which is the most recent one in a long-running mix series of mine dedicated to the sound of 90’s European hard trance, which is a type of music I fell in love with as a teenager. ‘Trance’ has over the years come to mean a pretty wishy-washy, even cheesy, form of dance music, but this is the raw, uncut stuff. Very intense!

Thanks for reading, see you again next time!